---
format: "aidr-story-markdown/v1"
id: "8538b660bdca9b4645e8802217ab8437ad2959cb2d19ae2a09fe338817e94b25"
canonical_url: "https://aidr.today/8538b660?lang=en"
title: "SGLang Speeds Kimi K3 Inference by 20% on NVIDIA Vera Rubin"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-10T02:40:05.000Z"
category: "Frameworks"
topics: ["nvidia","inference","framework"]
source_urls: ["https://huggingnews.com/ai/sglang-speeds-kimi-k3-inference-by-20percent-on-nvidia-vera-rubin-e243b253","https://marketbrief.now/ai/sglang-speeds-kimi-k3-inference-by-20percent-on-nvidia-vera-rubin-e243b253"]
summary: "SGLang achieved up to 20% faster FP8 MLA at batch 1 with a 128K context for the Kimi K3 model on NVIDIA Vera Rubin hardware. Other gains include 20% faster KDA verification and a 16% speedup for batch 16."
---

# SGLang Speeds Kimi K3 Inference by 20% on NVIDIA Vera Rubin

> [Open the canonical story](<https://aidr.today/8538b660?lang=en>)

**Published:** 2026-10-10T02:40:05.000Z
**Category:** Frameworks
**Topics:** nvidia, inference, framework

## Summary

SGLang achieved up to 20% faster FP8 MLA at batch 1 with a 128K context for the Kimi K3 model on NVIDIA Vera Rubin hardware\. Other gains include 20% faster KDA verification and a 16% speedup for batch 16\.

## Sources

- [Story source](<https://huggingnews.com/ai/sglang-speeds-kimi-k3-inference-by-20percent-on-nvidia-vera-rubin-e243b253>)
- [Story source](<https://marketbrief.now/ai/sglang-speeds-kimi-k3-inference-by-20percent-on-nvidia-vera-rubin-e243b253>)